The amygdala primarily evolved as a danger detector that regulates emotional behaviours and learning. However, it is also engaged in stress responses as well as olfactory/pheromonal and reproductive functions. All of these functions are processed by a set of nuclei which are derived from different telencephalic sources (pallial and subpallial) and have a unique cellular structure and specific connections. It is unclear how these individual anatomical and functional units evolved to fit the amygdala to the specific needs of various mammals. Thus, this study provides quantitative data regarding volumes, neuron density and neuron numbers in the main amygdala nuclei of the common shrew, guinea pig, rabbit, fox and pig - species from across the mammalian phylogeny which differ in brain complexity and ecology. The results show that the volume of the amygdala and its individual nuclei scale with negative allometry relative to brain mass (an allometric coefficient below one). However, in relation to the whole amygdala volume, volumes and volumetric percentages of the lateral (LA) and basomedial (BM) nuclei scale with positive allometry, for the medial (ME) and lateral olfactory tract (NLOT) nuclei these parameters scale with negative allometry while the values of these parameters for the basolateral (BL), central (CE) and cortical (CO) nuclei scale with isometry. Moreover, density of neurons scales with strong negative allometry relative to both brain mass and amygdala volume with values of allometric coefficient below zero across studied species. This value for BL is significantly lower than that for the whole amygdala, for ME it is significantly higher while values for NLOT, CE, CO, LA and BM are quite similar to the value for whole amygdala. Finally, neuron numbers in the whole amygdala and its individual nuclei scale with negative allometry in relation to brain mass. However, in relation to the number of neurons in the whole amygdala, neuron numbers and percentages of neurons for LA and BM scale with positive allometry, for BL and NLOT they scale with negative allometry while the values of these parameters for CE, CO and ME scale with isometry. In conclusion, all of these results indicate that each of the nuclei studied displays a different and unique pattern of evolution in relation to brain mass or the whole amygdala volume. These patterns do not match with the various classical concepts of amygdala parcellation; however, in some way, they reflect diversity revealed by the expression of homeobox genes in various embryological studies.